Re: updated WIP: arrays of composites

Tom Lane wrote:
>
> There is *tons* of legacy code that uses _foo, mainly because there was
> a time when we didn't support the [] notation in a lot of places where
> types can be named.  There still are some places, in fact:
>
> regression=# alter type widget[] set schema public;
> ERROR:  syntax error at or near "["
> LINE 1: alter type widget[] set schema public;
>                          ^
> regression=# alter type _widget set schema public;
> ERROR:  cannot alter array type widget[]
> HINT:  You can alter type widget, which will alter the array type as well.
> regression=#
>
> That particular one may not need fixed (anymore) but the real problem is
> the torches-and-pitchforks session that will ensue if we break legacy
> code for no reason beyond cosmetics.
>
> IIRC some of the contrib modules still have instances of _foo in
> their SQL scripts.
>

Then I think we need to work out a way to make pg_dump smart enough to
dump things in the right order.

Can we perhaps explicitly deprecate using the type name to refer to
array types?

cheers

andrew